grouping-noisy-errors
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from exception messages and stack traces, which could be manipulated by external actors to influence the agent's behavior.
- Ingestion points: The skill retrieves sampled
$exceptionevents, including stack traces and messages, via theposthog:query-error-tracking-issue-eventstool (referenced in SKILL.md). - Boundary markers: The instructions lack explicit requirements for the agent to wrap error content in distinct delimiters or markers to prevent the agent from accidentally obeying instructions embedded within the error data.
- Capability inventory: The skill possesses write capabilities including merging issues (
posthog:error-tracking-issues-merge-create) and creating grouping rules (posthog:error-tracking-grouping-rules-create), which could be misused if the agent is successfully injected. - Sanitization: There is no mention of programmatic sanitization or filtering of the error messages before they are processed by the agent; however, the skill provides a strict comparison checklist (matching SDK, exception type, and top stack frame) which provides some mitigation against accidental grouping of unrelated data.
Audit Metadata